Multi Agent Research Team (MART)

Academic and industrial research processes are still time-consuming, concerning manual data collection, summarization, and contextual integration. This article presents MART (Multi-Agent Research Team), An AI system that facilitates end-to-end research work by a collaborative multi-agent system.MART employs autonomous agents, large language models (LLMs), and multimodal data processing to improve query enrichment, real-time data collection, summarization, and context analysis. The system is user-friendly uploaded files (DOCX, PDFs) and pics, parsed and processed by GPT-4o and LLaMA 3.3 70B. An iterative summarization loops-enabled context-aware retrieval vector database (FAISS) facilitates high quality outputs. MART's full-stack web implementation includes user authentication, history tracking, and real time visualization, thus offering a scalable solution for dynamic research needs. By bridging gaps in existing tools, MART shows enormous improvements in research automation, accuracy, and individualization to the user.